Background: Despite the effectiveness of a two-stage exchange protocol for the treatment of deep periprosthetic infection, infection can persist after resection arthroplasty and treatment with antibiotics, leading to a failed second-stage reconstruction. Intraoperative analysis of frozen sections has been shown to have a high sensitivity and specificity for the identification of infection at the time of revision arthroplasty; however, the usefulness of this test at the time of reoperation after resection arthroplasty and treatment with antibiotics is, to our knowledge, unknown.Methods: The medical records of sixty-four consecutive patients who had had a resection arthroplasty of either the knee (thirty-three patients) or the hip (thirty-one patients) and had had intraoperative analysis of frozen sections of periprosthetic tissue obtained at the time of a second-stage operation were reviewed. The mean interval between the resection arthroplasty and the attempted reimplantation was nineteen weeks. The results of the intraoperative analysis of the frozen sections were compared with those of analysis of permanent histological sections of the same tissues and with those of intraoperative cultures of specimens obtained from within the joint. The findings of the analyses of the frozen sections and the permanent histological sections were considered to be consistent with acute inflammation and infection if a mean of ten polymorphonuclear leukocytes or more per high-power field (forty times magnification) were seen in the five most cellular areas.Results: The intraoperative frozen sections of the specimens from two patients (one of whom was considered to have a persistent infection) met the criteria for acute inflammation. Four patients were considered to have a persistent infection on the basis of positive intraoperative cultures or permanent histological sections. Overall, intraoperative analysis of frozen sections at the time of reimplantation after resection arthroplasty had a sensitivity of 25 percent (detection of one of four persistent infections), a specificity of 98 percent, a positive predictive value of 50 percent (one of two), a negative predictive value of 95 percent, and an accuracy of 94 percent.Conclusions: A negative finding on intraoperative analysis of frozen sections has a high predictive value with regard to ruling out the presence of infection; however, the sensitivity of the test for the detection of persistent infection is poor.